Strange behavior I just noticed with Overwatch 2

[gameinn]

Since Reflex uses a dynamic FPS limit to prevent GPU usage from maxing, something in it's behavior may briefly engage the dynamic limit during the larger frametime spikes/variances, causing what I assume would be a relatively benign effect if you're system is already momentarily dropping to 100 FPS during said variance (which is usually too brief to be picked up by overlays like Afterburner due to polling report limitations)

The thing is I'm running a 5950x with a 3070. This config can easily keep above 250 fps regardless of scenario and in the practice range it's definitely 350 fps+ esepcially when I'm just standing still. I can't see how under any circumstance my system would have 100.

Can anyone confirm on their end?

Edit: Ok I got lucky. As I was looking at the SIM, received a discord notification. It went up as that appeared on screen. Switched off discord and now the most I saw was 6.1ms. Turned reflex off and most I saw was 6.0ms. I still don't know what is accounting for the 0.2-0.3 difference but it seems to have been Discord overlay as the culprit. Is it stupid to say the discord overlay is likely internally running like 60 fps or something so when it pops up reflex is fighting it? idk but atleast that's been addressed.

[jorimt]

The thing is I'm running a 5950x with a 3070. This config can easily keep above 250 fps regardless of scenario and in the practice range it's definitely 350 fps+ esepcially when I'm just standing still. I can't see how under any circumstance my system would have 100.

Momentary frametime variances don't always affect average framerate. If you say OW reported a maximum frametime of 10ms, it means at least one frame in that session took that long to complete, even if your average framerate visually remained the same.

A frame with a slower-than-average frametime here and there will not always affect average reported framerates so long as there are enough frames with frametimes that reach that reported average consistently each second.

--------

As per entry #2 of my Closing FAQ:
https://blurbusters.com/gsync/gsync101- ... ttings/15/

“Frametime” denotes how long a single frame takes to render. “Framerate” is the totaled average of each frame’s render time within a one second period.

At 144Hz, a single frame takes 6.9ms to display (the number of which depends on the max refresh rate of the display, see here), so if the framerate is 144 per second, then the average frametime of 144 FPS is 6.9ms per frame.

In reality, however, frametime from frame to frame varies, so just because an average framerate of 144 per second has an average frametime of 6.9ms per frame, doesn’t mean all 144 of those frames in each second amount to an exact 6.9ms per; one frame could render in 10ms, the next could render in 6ms, but at the end of each second, enough will hit the 6.9ms render target to average 144 FPS per.

--------

Whenever there are individual frames that deviate enough from the currently reported average framerate, it can be experienced as stutter instead.

Edit: Ok I got lucky. As I was looking at the SIM, received a discord notification. It went up as that appeared on screen. Switched off discord and now the most I saw was 6.1ms. Turned reflex off and most I saw was 6.0ms. I still don't know what is accounting for the 0.2-0.3 difference but it seems to have been Discord overlay as the culprit. Is it stupid to say the discord overlay is likely internally running like 60 fps or something so when it pops up reflex is fighting it? idk but atleast that's been addressed.

Any background process (be it an overlay or program) can interrupt the running game process enough to affect frametime performance. That's obviously what you were seeing then.

[gameinn]

Any background process (be it an overlay or program) can interrupt the running game process enough to affect frametime performance. That's obviously what you were seeing then.

I know this is likely beyond your knowledge since it's pretty niche but do you know if tightening timings of RAM will give slightly less sporadic frametime or doesn't really matter?

Also is there a graph or something that shows what frametimes a specific fps should give? e.g. as you said earlier 157 = 6.4 and 164 = 6.1

[jorimt]

I know this is likely beyond your knowledge since it's pretty niche but do you know if tightening timings of RAM will give slightly less sporadic frametime or doesn't really matter?

Tighter RAM timings can (but not always depending on the RAM, RAM tuning, platform, remaining system specs, and test game) result in higher minimum framerates, which ultimately can translate to less potential frametime variances and spikes, assuming the RAM (if overclocked) remains stable, of course.

Also is there a graph or something that shows what frametimes a specific fps should give? e.g. as you said earlier 157 = 6.4 and 164 = 6.1

The simplest formula to calculate the frametime of an average framerate is 1000 / x, thus 1000 / 60 (FPS) = 16.66666666666667 (ms). We then remove all but the first decimal point for a friendlier looking number. I.E. 60 FPS = 16.6ms frametime per frame, and so on...

There used to be a convenient calculator on a website that has since been removed (web archive version):
https://web.archive.org/web/20220528040 ... lator-fps/

There's also another live calculator for the same purpose I found here:
https://fpstoms.com/

[jabbo]

Jorimt, since you have OW2 now would you mind testing another odd behavior I noticed? So OW engine has an interesting indicator on its in-game FPS monitor where it can show dots at the end which mean different things. Per Blizzard, 1 dot after your fps means SIM bound (which I understand to be optimal operation where only limitation is game engine), 2 dots means GPU render thread full/laggy, and 3 means a problem in your render thread (which I assume is CPU?). You can google and see some Blizzard responses on this which is what I did.

The interesting behavior is this:
1) if you enable in-game vsync, no matter what else is enabled, the dots disappear entirely. I can't tell if this is actually zero dots thus best behavior OR (and most likely) just a visual bug.
2) regardless, if I turn off in-game vsync and enable NVCP vsync (note I also use Gsync) then the fps display will always show 2 dots. With Vsync off, everywhere, but Gsync on, if I cap framerate and enable reflex or reduce buffering the dots go back to 1.

So, I guess my question is, could you replicate this? And if you can, what do you think about the nvcp vsync option causing 2 dots to appear as if there is a GPU render/thread issue? Could it be this game, with gsync, its optimal to only use in-game vsync or simply no vsync at all?

[jorimt]

Jorimt, since you have OW2 now would you mind testing another odd behavior I noticed?

Turns out I technically always had it, since I already owned OW. I just didn't know they had directly replaced OW with OW2 until I tried to reinstall OW.

So OW engine has an interesting indicator on its in-game FPS monitor where it can show dots at the end which mean different things. Per Blizzard, 1 dot after your fps means SIM bound (which I understand to be optimal operation where only limitation is game engine), 2 dots means GPU render thread full/laggy, and 3 means a problem in your render thread (which I assume is CPU?). You can google and see some Blizzard responses on this which is what I did.

The interesting behavior is this:
1) if you enable in-game vsync, no matter what else is enabled, the dots disappear entirely. I can't tell if this is actually zero dots thus best behavior OR (and most likely) just a visual bug.
2) regardless, if I turn off in-game vsync and enable NVCP vsync (note I also use Gsync) then the fps display will always show 2 dots. With Vsync off, everywhere, but Gsync on, if I cap framerate and enable reflex or reduce buffering the dots go back to 1.

So, I guess my question is, could you replicate this?

I've bookmarked this post and will look into it if/when I get the time, but no ETA. My blind guess is it means little to nothing and has little to no impact on anything important.

what do you think about the nvcp vsync option causing 2 dots to appear as if there is a GPU render/thread issue? Could it be this game, with gsync, its optimal to only use in-game vsync or simply no vsync at all?

I think devs should stop adding vague readouts that induce paranoia in their playerbase.

Again, I don't think what you're seeing means what you probably think it means. All V-SYNC does (with or without G-SYNC) is time the beginning of each frame scan to the VBLANK signal at the start of each scanout cycle to prevent tearing.